diff --git a/apps/worker/video_processing/render_adapter.py b/apps/worker/video_processing/render_adapter.py
index 4f5cb5444..c52f3b2d0 100755
--- a/apps/worker/video_processing/render_adapter.py
+++ b/apps/worker/video_processing/render_adapter.py
@@ -575,12 +575,13 @@ class RenderAdapter:
 
         self._report_progress(progress_cb, 90.0, "生成封面缩略图")
 
-        # 6. 生成封面缩略图
+        # 6. 生成封面缩略图（结果通过 RenderAdapterResult.thumbnail_url 返回给调用方）
         thumbnail_url = ""
         try:
             from video_processing.thumbnail_generator import generate_and_upload_thumbnail
 
-            thumb_storage_key = f"rendered/{plan_id}/thumbnail.jpg"
+            # job_id 是 _do_render 方法的参数（参见方法签名）
+            thumb_storage_key = f"rendered/{plan_id}/thumbnails/{job_id}.jpg"
             thumbnail_url = generate_and_upload_thumbnail(str(result.output_path), thumb_storage_key)
         except Exception as thumb_err:
             logger.warning(
diff --git a/apps/worker/worker_app/tasks/generation.py b/apps/worker/worker_app/tasks/generation.py
index 8dbbde32b..7915bd075 100644
--- a/apps/worker/worker_app/tasks/generation.py
+++ b/apps/worker/worker_app/tasks/generation.py
@@ -389,6 +389,7 @@ def _upload_and_record(
     editing_mode,
     user_id: str = "",
     video_name: str = "",
+    thumbnail_url: str = "",
 ) -> tuple[str, float, int, int]:
     """上传 OSS、创建视频记录并查重。
 
@@ -447,6 +448,7 @@ def _upload_and_record(
             mode=editing_mode.value,
             session=dedup_session,
             name=video_name,
+            thumbnail_url=thumbnail_url,
         )
     finally:
         dedup_session.close()
@@ -535,11 +537,11 @@ def _render_from_edit_plan(
     task_id: str,
     source_edit_plan_id: str,
     task_info: dict,
-) -> tuple[Path, float, list[dict] | None, str | None, str | None]:
+) -> tuple[Path, float, list[dict] | None, str | None, str | None, str]:
     """从 EditPlan 数据库记录直接渲染（不再内存重建clips）。
 
     Returns:
-        (output_path, render_duration, cover_candidates, voiceover_path, temp_dir)
+        (output_path, render_duration, cover_candidates, voiceover_path, temp_dir, thumbnail_url)
     """
     from video_processing.render_adapter import RenderAdapter
     from worker_app.db import SessionLocal
@@ -580,7 +582,14 @@ def _render_from_edit_plan(
         cover_candidates = getattr(result, "cover_candidates", None)
         render_temp_dir = getattr(result, "temp_dir", None)
 
-        return output_path, result.duration, cover_candidates, voiceover_path, render_temp_dir
+        return (
+            output_path,
+            result.duration,
+            cover_candidates,
+            voiceover_path,
+            render_temp_dir,
+            result.thumbnail_url or "",
+        )
     finally:
         db.close()
 
@@ -692,7 +701,7 @@ def generate_video(self, task_id: str) -> dict:
                     gen_task.append_log("渲染模式", "从草稿数据渲染（与预览一致）")
                     _flush_logs(task_id, gen_task)
 
-                output_path, render_duration, cover_candidates, voiceover_tmp_path, render_temp_dir = (
+                output_path, render_duration, cover_candidates, voiceover_tmp_path, render_temp_dir, thumbnail_url = (
                     _render_from_edit_plan(
                         task_id=task_id,
                         source_edit_plan_id=source_edit_plan_id,
@@ -717,6 +726,7 @@ def generate_video(self, task_id: str) -> dict:
                         editing_mode=editing_mode,
                         user_id=user_id,
                         video_name=task_info.get("video_title", ""),
+                        thumbnail_url=thumbnail_url,
                     )
 
                     if gen_task: